Possible Issue with Calculating Guaranteed Loss

[naterush]

This is a more-game-theoretic-issue than bug-in-the-code. I’m also not totally positive about it but would love to hear your thoughts.

Assuming: At some point in the future, the total supply of Ether will be relatively constant (reduced issuance, the possible burn of tx fees, loss from lost keys/death, etc).

Early adopters w/in curation market make a "profit". This means that some of the late adopters (say, at least, the very last adopter), will necessarily lose money.

Thus, if you knew you were the last adopter, you would choose to not invest - as you would be guaranteed to lose money.

Furthermore, as there is a limited supply of Ether, if all other Ether was invested in the market, the last un-invested Ether holder could be sure they would lose money, and would thus choose to not invest.

As the last user would never choose to invest, thus the person before them would never choose to invest (as they can be sure, by this argument, there will be no later adopters to invest and give them a profit).

The argument can be applied recursively — which seems to point out that it is not a stable equilibrium to invest in the first place.

Essentially, the argument is that a user who is thinking about putting money into the market can say “If the rest of all free Ether was invested in this contract, would I end up making a profit, given the amount of Ether invested as of now?” — and come to the conclusion that it does not make sense to invest.

I simplified the argument by saying “the last adopter,” but from some (possibly wrong) calculations, its seems mores like the last >75% of adopters would not make their money back. But as long is there is one (which there must be if early adopters get rewarded), I think the argument holds.

I might be totally off-base here, but let me know what you think.

[simondlr]

Fascinating point. I always like to push designs to the extreme to discover potential flaws.

So, the issue is mainly that ETH could dry up, right? eg, there's no ability anymore for anyone else to get in? And thus, knowing that, the last person wouldn't get in, and ad infinitum.

I think it's an interesting point. However, I don't think the market would be as rational.

1. How would you know that you are putting the last ETH in?
2. If you know that the last ETH would possibly be inserted, and you are second in line, how would you know that the person after you would put in the rest? How do you know you will be second last? Or third last? Or fourth last? etc
3. Getting benefit from this is not just about the increase in token value. The first and foremost value prop is to get the coordination benefits. So even if you are somehow the last buyer, you are still wanting to get in, due to being able partake in the community.
4. Additionally, ETH is currently just a way to have a "cost" to enter. In the future it can be other things as well, like a USD stable token, or a promise to fullfill something useful to the community and it being backed by some reputation system.

What do you think?

[naterush]

You're probably right that the market would be nowhere near as rational as I'm making it out to be. :)

However, I think there might be some other concerns with how buy-ins/payouts (B/P) are calculated.

1. Due to B/P, early investors are as likely to be squatters as they are interested curators.

Argument: The first investor cannot lose money. Therefore, it is the optimal strategy to invest first no matter what, even if you do not intend to curate. (Capitol lockup cost is negligible, as a user can withdraw at any point).

Furthermore, I calculated that the ~20-30% of the earliest investors will make a profit. This also illustrates that it will be easier for a group of early investors to make money of a pump-and-dump than later ones, as earlier ones have to convince a much smaller amount of ETH to join to make a profit. Thus they may have an incentive to do so -- rather than actually curate.

This early-bad-curator issue is likely mitigated by having multiple curation markets allowed for each topic (#whitepaper), but I wonder if it might create a not-so-positive-air around curation markets as a whole.

2. Due to B/P, later investors have much less incentive to curate.

Argument: Later investors have significantly lower potential profit, and also stand to profit less from each additional ETH invested into the market, as they own a smaller share of coins.

As a result, they will have less incentive to curate, as an equal unit of curation from them and from an early investor will benefit an earlier investor more.

The combination of both of these makes it seem like the current B/P might not do the best job of incentivizing good curation; the early adopters might not be the best curators, and the later adopters may be less incentivized to do so.

I obviously make a ton of assumptions in arguments above, but I wonder if it would be possible to design a better cost/payout mechanism that might incentivize better curation. Would love to hear your thoughts :)

[naterush]

Here’s a possible design scheme that might better incentivize curation while discouraging bad behavior.

When a user buys in, X% of their tokens minted are locked and not withdraw-able or bondable.

Furthermore, when users buy in, they can claim some set of curators that were responsible for curating content that caused them to join.

If you are in set of curators a user claims when buying in, or your tokens are bonded to that curator, then some of the locked X% of tokens are unlocked according to how much ether this new user invested.

This scheme could make it so users only make a profit in the case where the total amount of Ether they encouraged to join is greater than the Ether they are withdrawing from the market.

Thus, users will only make a profit if they effectively curate, or delegate their coins to someone else to curate (it also might make sense to give curators more of the reward). I believe this would disincentivizes buying into for any other reason than being a good curator - while also incentivizing good curation once in the market.

Would love to hear your thoughts. Also, sorry for the walls of text. Curation markets are just too cool :).

[pgorsira]

Furthermore, when users buy in, they can claim some set of curators that were responsible for curating content that caused them to join.

In practice, I would imagine that behavior would be more similar to the existing curator-bonding process in that new participants would opt to select those curators that show the most promise generally rather than those that specifically "recruited" them. Seems like with the proper tools to take a look at all signals within the curation market (that is, even if you are brand new to this market) new participants would view this bonding process through the same lens as existing participants in terms of which curator(s) to bond to.

I'm not sure of any definitive mechanism for "forcing" the selection of the proper referring curator upon a new participant joining the curation market. Then again, I'm not so sure that it matters.

[MagRelo]

Super interesting discussion! I've been thinking along the same lines, and I think that one really cool opportunity with continuous tokens is that you can calculate the 'buy-in' price and the 'cash-out' price on-the-fly, and so you can influence the dynamics of the network as you go.

So if you're worried about early tokens dumping you can just design a 'cash-out' price function that would consider the age of the token and the volume of the trade and return a discounted price that would discourage that behavior.

I refactored CurationMarkets.sol a bit to highlight this (not tested, just for illustration) https://github.com/MagRelo/curationmarkets/blob/master/CurationMarkets.sol#L119

thoughts? cheers!

[simondlr]

Thanks for the comments guys! Sorry for the late reply!

I wonder if it would be possible to design a better cost/payout mechanism that might incentivize better curation locking scheme

@naterush very very interesting proposal. I like that.

1. This changes UX a bit, but I think UX is a faaarr off worry atm. Users joining have a new burden of evaluating curators.

2. To what extent does make things potentially game-able in the design? Or rather, could it side-stepped? Do you see any issues with this? Need to do some deep diving here.

I would imagine that behavior would be more similar to the existing curator-bonding process in that new participants would opt to select those curators that show the most promise generally rather than those that specifically "recruited" them

@pgorsira ah, good point. Because good curators attracted their attention to start with. So it might not be that different from a UX perspective after all.

@MagRelo

Nice, yeah. It allows for some interesting emergent properties (like ranking curation markets and potentially attaching prediction markets to the prices). eg -> Will a topic be popular in June? Now you have an on-chain metric for that. :)